Find value in the AI stack D3

Find value in the AI stack As AI’s investment opportunity broadens, finding value is becoming more important – and more difficult. Our framework helps investors navigate Key takeaways the AI ecosystem layer by layer, testing where the investment case still holds, where bottlenecks are emerging and where pricing power and • Our framework applies testable profits may concentrate. questions to AI’s “five-layer cake” – applications, models, infrastructure, Investing in artificial intelligence (AI) can feel daunting. The ecosystem chips and energy – to assess if and is broad, spanning vast data centres through to the applications used by where the investment case remains businesses and governments. So how can investors capture existing and intact. emerging opportunities and seek to maximise returns while managing risk? • This approach helps determine where, and how much, capital to Rather than viewing AI as a single investment theme, we assess each stage allocate across overall AI exposure of the value chain separately. This helps identify where the investment case and specific layers of the ecosystem. remains correct, where risks are emerging, and where economic returns are shifting. • Our analysis shows that leading indicators – including adoption While many AI investment analyses focus on how the technology is built, trends, infrastructure build-out and our approach focuses on where the investment case remains strongest, semiconductor demand – can help where value can be created, and which parts of the ecosystem are to reveal shifts before they show up strongly positioned to outperform – and those that may be constrained by in earnings. bottlenecks. • As AI adoption broadens, leadership may rotate across the value chain, with energy availability and grid capacity becoming potential critical constraints.

Gregor MA Hirt Martin Lubojanski Sebastian Lukas CIO Multi Asset Portfolio Manager Portfolio Manager ALLIANZGI.COM

Testing the AI investment thesis Nvidia coined the term “five-layer cake” to describe the AI ecosystem as a comprehensive, full-stack industrial Our investment framework assesses the AI investment infrastructure with the following sequential layers: opportunity through a series of testable questions spanning the entire value chain. It has three distinguishing 1. Applications – the AI products and services that solve features. problems for users.

First, it is diagnostic rather than predictive. The 2. Model – the systems trained to understand language, framework’s objective is to evaluate the health of the images, video, biology and more. AI investment thesis through a sequence of falsifiable 3. Infrastructure – the data centres, networking questions. We use data from multiple perspectives to equipment, cooling systems, power connections and distinguish meaningful AI developments from hype. physical facilities needed to operate AI at scale. Second, it focuses on leading rather than lagging 4. Chips – the graphics processing units (GPU) and central indicators. A diverse range of operational metrics – such processing units (CPU), memory chips and advanced as the speed of AI adoption by companies or the breadth semiconductor components that make AI possible. of employee usage– can point to where the market is 5. Energy – the power needed to run data centres. heading before quarterly earnings confirm the trend. We use this structure to link each layer to a testable Third, it acknowledges that economic profit pools shift investment question, supported by objective indicators. over time. AI adoption defines the size of the opportunity, Rather than forecasting quarterly earnings or trying to while bottlenecks and competitive advantages determine identify the next short-term winner, we assess whether the who benefits. For example, if high-bandwidth memory structural AI investment thesis remains intact and where becomes a limitation, companies in the sector may be economic rents – excess profits earned from controlling able to raise prices, supporting stronger earnings and, scarce digital assets – are emerging as the industry evolves. potentially, higher share prices. We apply that lens across the AI ecosystem (see Exhibit 1).